Microsoft Has Built a Linux Distro

jbernardo writes: Microsoft has built a Linux distro, and is using it for their Azure data centers. From their blog post: "It is a cross-platform modular operating system for data center networking built on Linux." Apparently, the existing SDN (Software Defined Network) implementations didn't fit Microsoft's plans for the ACS (Azure Cloud Switch), so they decided to roll their own infrastructure. No explanation why they settled on Linux, though — could it be that there is no Windows variant that would fit the bill? In other news, Lucifer has been heard complaining of the sudden cold.

Comments

MS uses what works

[danbuter]

It's in-house and they aren't trying to sell it. No reason not to use Linux.

Re:MS uses what works

[neilo_1701D]

Exactly.

I'm sure that given time and money, there could be a Windows variant that did the job. But that isn't MS's focus. Here in the Microsoft Dynamics consulting world, Azure is what is being pushed hard for all the latest enterprise systems (CRM, ERP). Microsoft makes it's money from Azure and everything that runs on top of that. This stuff is nothing to them.

Re:MS uses what works

[bug1]

No reason not to use Linux.

Except that its un-American, and causes cancer...

Re:MS uses what works

[Penguinisto]

Speaking of which, I wonder how long it will take for Linux to 'metastasize' within the organization?

First, it fulfills a couple of roles here and there in MSFT. Next, they have to make their own in-house distro. Next, they discover that it's kind of useful for a few internal roles within a few internal departments (esp. budget-starved ones). Next...?

Slowly, surely... ?

Re:MS uses what works

[mwvdlee]

The big difference is that the TCP/IP stack used a BSD license but Linux has a GPL license.
You can use BSD code, add a license notice (on original BSD license) and be done.
If MS is offering the Linux distro to it's users, then it must make available it's Linux distro's code too.

Re:MS uses what works

[number6x]

No reason, other than Microsoft's own policy of 'Eating Its own dog food'.

This policy was in place in order to force Microsoft to develop its own solutions from within its own software, in order to force their own software to become continually better and better.

Of course Microsoft doesn't build their own chairs and desks for their offices, so where do you draw the line between the dog food policy and using other's products for solutions instead of their own? Office furniture is a no-brainer, Microsoft has no dog food to eat. Enterprise level RDBMS data bases would be another, as SQL Server is not really in the same class as Oracle or DB2. Linux, however is different. Linux is a general use OS for Intel (and other) based computers. Windows is a general use OS for Intel based computers. This is a pretty significant cross over. Anything Linux can do, Windows should be able to do. Not improving Windows to be able to match or beat Linux at something is definitely choosing to eat someone else's dog food.

It may show that Microsoft is shedding some of their traditional 'rules' in order to transform the company and create a new Microsoft.

Wrong choice

[BitZtream]

This was just a bad choice. If they wanted a proper software defined network, they'd have selected FreeBSD since it has the fastest, most compact networking stack in the world and its well known/accepted fact by anyone who does high-end networking, hence why Microsoft ALREADY has a fuck ton of FreeBSD installs on their core network labeled ... Juniper Networks ... or F5 ... or any of the other ones.

Someone deserves to get fired for this. Not because they picked Linux, but because Linux simply wasn't the right choice in any way shape or form as every other major company doing networking has illustrated.

Re:Wrong choice

[Penguinisto]

Not so fast - after all, guess what Cisco chained into their Nexus line of switches? (NX-OS is not using a FBSD kernel, after all.)

It's not that FBSD is failed or failing, but because Linux has a much bigger mindshare nowadays, which means you can more easily get the real esoteric and custom bits for your needs, especially without having to write it all yourself.

Yes, I know FBSD has linux compatibility and stuff, but that's not the point.

Re:Wrong choice

[coolmoe2]

Sooo tired of FBSD blowhards claiming that it is the right choice for everything. Linux has a very large developer base and is mature enough to give all of the BSD's a run for their money. Linux is everywhere now days from the data center to your smartphone. Your claims that somebody should be fired for using it is just plain childish.

Re:Wrong choice

[Burdell]

I suggest you look up how SDN switches and Juniper routers work. In neither case is the commodity OS used in the forwarding path; it is just a control-plane OS, and the performance of the control-plane OS's network stack has no bearing on traffic performance. Whether FreeBSD's network stack is "better" than Linux's is debatable (I thought all the BSD-heads "knew" that OpenBSD's network stack was the best, not FreeBSD), but it has no relevance here.

Re: Wrong choice

[tomhath]

It will suffice, hence he should have only used "hence" is why.

Clever

[Anonymous Coward]

By launching their own distro, Microsoft has figured a way to grab Linux for free and make it another money-making machine for them. Now, this is ironic. Well played, I have to admit.

This proves Linux sucks

[Anonymous Coward]

The Total Cost of Ownership is so high, that only a company as rich as Microsoft can use it for their own business.

Re:Satan

[YrWrstNtmr]

No, Satan runs BeOSelbub.

Not MS's first Linux

[mschaffer]

This is not the first Linux released by Microsoft.
The first one was released in 2003. http://www.mslinux.org/
It was released under GPL (Gates Private License).

So where can I download MS-Linux?

[Spy Handler]

I can't find it on torrent sites.

Re:Not the first time...

[moehoward]

Microsoft had XENIX back in the late '80s and early '90s. And, it was available to anyone. I supported many customers on it with our software and really enjoyed working with it. PC people couldn't believe that you could run a 386 or 486 and support multiple users at the same time with cheap dump terminals. And as I type this from a Linux-based Chromebook, I couldn't be happier that *NIX is not only eating Microsoft's lunch, but it is also being served for lunch at Microsoft.

We all saw this coming. And we know where it is all going...

If it doesn't use systemd, I'd like to use it.

[Anonymous Coward]

I'm not joking around here. If Microsoft put out a Linux distro that didn't use systemd, with some guarantee that it never would, I'd very much consider using it. It sounds absolutely crazy, but things have gotten so fucked up in the Linux ecosystem lately that the thought of Microsoft putting out the best Linux distro has actually become plausible.

Re: If it doesn't use systemd, I'd like to use it.

[EmeraldBot]

The Slackware FAQ still talks about SoundBlaster 16 and old CD-ROM drives. It gives a strong impression that this is not a distro for modern times.

Well evidently, since you haven't even bothered to look at it beyond the home page, you clearly haven't been pressured enough into bothering to do any real research. Slackware is literally just a vanilla Linux kernel and some prebuilt packages of popular programs - that's all it is. No custom this, no custom that, no preset defaults. If the plain Linux kernel isn't stable, reliable, trustworthy and has a large community, then I don't know why you'd think Debian would have that.

Of course, you're actually right in some ways. While it's designed to be simple to modify for what you wish it to be, it really excels at being a personal OS for a single user, one who can mold it into whatever they want. In particular though, it's not suitable for enterprise use, or at least not without some serious custom modding and testing. The packages, while stable, are fairly new, the prebuilt package repository is fairly small (and the unofficial Slackbuilds isn't stable at all), and the whole package managment system in general doesn't really scale well. The difference is, I actually used it heavily a couple years back, before I switched my peronal workstation to FreeBSD, and I found its weak points (and strong ones) through heavy daily interaction. If you're willing to dismiss it simply because of what it looks like alone, you're clearly not the target audience. So go back to Debian, apt-get purge that Systemd, and put back on your beloved sysvinit. Go ahead, it'll be the same as before, it won't have been cursed by thy evil foe then.

But if you can't be bothered to do any sort of serious study whatsoever, then please quit whining like a three year old, and don't bite off someone's head when they gave you a well-intentioned and helpful reccommendation.

Re:If it doesn't use systemd, I'd like to use it.

[Anonymous Coward]

We will never use use systemd. We have replaced it by svhost.exe (Microsoft insider).

Re:If it doesn't use systemd, I'd like to use it.

[sparkeyjames]

Bullcrap. I've built a few email servers and firewall boxes running Slackware. A few even have nine nines uptime. They are rock solid and damn near impenetrable. Your only argument is that they may not be great for cloud services and even that is debatable. Do you like the taste of foot?

Do you blame them?

[Trailer Trash]

I mean, Linux is just full of their patented inventions - hell, they practically wrote the whole thing! They should use it, and proudly!

Blue Screen of Death

[jfdavis668]

The major change is adding the Blue Screen of Death, just to make everyone comfortable with using Linux.

Re:Satan

[Tx]

Linux? Pah! There's a reason why Microsoft owns WindowsPowersHell.org.

Systemd

[snookiex]

Cool. Now they will bundle Clippy into Systemd.

Re:Systemd

[halivar]

No, that's not Clippy, that's Grubby.

So not publically not eating your own dog food

[Viol8]

..is fine, right? When you're trying to sell Windows to the public as a one size fits all OS yet its apparently not good enough to run the network of their own Premier cloud service thats not a problem?

Give me a break, this has embarrassing U-turn written all over it.

Re:So not publically not eating your own dog food

[neilo_1701D]

Your comment would be correct 5 years ago.

Now, cloud services are the thing.

As an example, the premier ERP solution that Microsoft has, Dynamics AX, is currently totally tied to Windows. The next version, AX 7, changes the game completely. The rich client - the bit the user interacts with - is gone, replaced with a browser-agnostic UI (sporting a Windows 8 Start screen look-and-feel, but that's another story). The server and database components are now running on Azure. Windows has effectively vanished from the equation. And this the flagship ERP application.

For another example, look at Microsoft Office.

Microsoft is no longer the company that makes Windows and defends the Windows franchise; it's now services, services, services, and Windows with stand-alone Office etc.

Re:So not publically not eating your own dog food

[steveha]

I have worked at Microsoft, and they are all about eating their own dog food. Everyone at Microsoft uses Microsoft products for everything.

And, let me remind you of the fiasco where Microsoft bought Hotmail and switched its servers from UNIX (FreeBSD on front-end servers and some Solaris database servers) to Windows. They had to throw more hardware at the operation and still had problems, but they did it, and they knew going in that they would have more problems with Windows.

But now we are talking about Azure. Microsoft is seriously going for market share in cloud hosting, and most of the customers they are trying to win over are already running their stuff on Linux. So it's not really that embarrassing for Azure to run on Linux... I attended the Linuxfest Northwest conference this year, and Microsoft Azure had a booth in the vendor room where they had signs saying "Microsoft <heart> Linux".

Also, Microsoft is going after the Docker market. They are whipping together something like Docker for running Windows server apps in the cloud, but meanwhile they are all in on supporting Linux Docker apps for Azure. They have ported the Docker admin tools to run on a Windows machine, so that people can control Docker from a Windows machine (while the Docker is still running on Linux, you understand).

Give me a break, this has embarrassing U-turn written all over it.

I disagree about the "embarrassing" part. Microsoft has, in the past, acted like it could control the industry. One reason it acted that way was that it used to succeed more often than not in actually controlling the industry. But it's far too late for Microsoft to kill Linux; they are going to have to co-exist with Linux forever now, and it's not embarrassing for them to act like it.

I remember, about seven years ago, seeing a video at Microsoft that showed a skinny kid on a skateboard as a visual metaphor for Linux. I was amused... did they really think they could convince IT guys to choose Windows over Linux just by sneering at Linux in a marketing video? The Microsoft that made that video could never make its own distro.

In recent years, Microsoft has not shown much ability to adapt. Look at how horrible their strategy was with portable music players and then with mobile devices. But now, the Azure guys are just doing whatever makes the most sense for them, and it is politically possible at Microsoft? That's actually a good omen for Microsoft's future; at least they are not denying reality as much as they used to.

Re:Not the first time...

[fnj]

Xenix was licenced from the then SCO to Microsoft :) It was therefore a rebadged SCO Unix it wasn't made by MS.

Just No. Microsoft licensed Unix v7 from AT&T in 1978 and announced in 1980 that they would make it available for 16 bit micros. Microsoft's license did not allow it to use the UNIX name, so they came up with Xenix as a name. Microsoft did not choose to sell it directly to end users. They licensed it to OEMs IBM, Intel, Tandy, Altos, SCO, and Siemens.

SCO was originally farting around with their own port of v7, Dynix. In 1982 they made a deal with Microsoft to jointly develop Xenix.